Chinese Journal of Aeronautics, (2016), 29(5): 1436–1444

Chinese Society of Aeronautics and Astronautics

& Beihang University
Chinese Journal of Aeronautics
cja@buaa.edu.cn
www.sciencedirect.com

Plastic deformation analysis and forming quality

prediction of tube NC bending
Lu Shiqiang a,*, Fang Jun a,b, Wang Kelu a

a
National Defense Key Disciplines Laboratory of Light Alloy Processing Science and Technology,
School of Aeronautical Manufacturing Engineering, Nanchang Hangkong University, Nanchang 330063, China
b
Jiangxi Key Laboratory of Surface Engineering, Jiangxi Science and Technology Normal University, Nanchang 330013, China

Received 2 November 2015; revised 12 January 2016; accepted 20 February 2016

Available online 26 August 2016

KEYWORDS Abstract Plane strain assumption and exponent hardening law are used to investigate the plastic
Forming quality; deformation in tube bending. Some theoretical formulae including stress, curvature radius of neu-
Plane strain; tral layer, angle of neutral layer deviation, bending moment, wall thickness variation and cross-
Tube NC bending; section distortion, are developed to explain the phenomena in tube bending and their magnitudes
Theoretical analysis; are also determined. During unloading process, the springback angle is deduced using the virtual
Virtual work principle work principle, and springback radius is also given according to the length of the neutral layer
which remains unchanged before and after springback. The theoretical formulae are validated by
the experimental results or the validated simulation results in literature, which can be used to
quickly predict the forming quality of tube numerical control (NC) bending.
Ó 2016 Production and hosting by Elsevier Ltd. on behalf of Chinese Society of Aeronautics and
Astronautics. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction and rolling bending, the numerical control bending (NC) is

As the ability in satisfying the demands of high strength/weight
ratio and impact energy absorption, bent tube parts have
attracted more and more applications in various high technol-
ogy industries, such as aviation, aerospace, automobile and
shipbuilding.1,2 Compared with other various bending meth-
ods such as press bending, stretch bending, pushing bending
* Corresponding author.
E-mail address: niatlusq@126.com (S. Lu).
Peer review under responsibility of Editorial Committee of CJA.
Production and hosting by Elsevier
the most universal method to produce the bent tubes. How-
ever, the effects of coupling multi-factors and lack of proved
theoretical instruction in tube NC bending result in the com-
plexity in forming mechanism, and multi-defects such as wrin-
kling, over thinning, cross-section distortion and springback.
Thus, great efforts have been conducted on the forming quality
prediction of tube bending by experimental, theoretical and
numerical approaches.
Wu et al.3 experimentally investigated the effects of the
temperature, bending velocity and grain size on the wall thick-
ness change, cross-section distortion and springback of AM30
magnesium alloy tube in rotary draw bending. Liu et al.4,5
experimentally obtained the effect of dies and process param-
eters on wall thickness distribution and cross-section distortion
of thin-walled rectangular 3A21 aluminum alloy tube in NC

http://dx.doi.org/10.1016/j.cja.2016.03.009
1000-9361 Ó 2016 Production and hosting by Elsevier Ltd. on behalf of Chinese Society of Aeronautics and Astronautics.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Plastic deformation analysis and forming quality prediction of tube NC bending 1437

bending. Tang6 employed the plastic-deformation theory to cross-section distortion, and springback angle/radius. The
deduce several bending related formulae such as stress distri- assumptions are as follows:
butions of bent tube, wall thickness variation, bending
moment and flattening. Al-Qureshi and Russo7 derived the (1) The uniform stress/strain distributions and the propor-
approximate equations of tube bending to predict springback tional loading are assumed.
and residual stress quantitatively with assumptions of ideal (2) The cross-section of tube remains in a plane during
elastic-plastic material, plane strain condition, absence of bending, and the shear deformation of cross-section is
defects and ‘‘Bauschinger effect”. E et al.8 derived an analytical negligible.
formulae for the prediction of the stress distributions, neutral (3) The cross-section deformation of tube is not considered,
layer shift, wall thickness change and cross-section distortion namely the circumferential strain is regarded as zero.
based on in-plane strain assumption in tube bending with (4) The stress-strain relation follows the exponent law as
exponential hardening law. Li et al.9 established the analytical
¼ K
en , where K is strength coefficient and n is harden-
r
prediction model for springback, which was considered tube ing exponent.
specifications and material properties, using the plastic- (5) The material of tube is incompressible during bending
deformation theory. The approximate equations were derived process.
to predict bending moment and springback of bent tube con-
sidering the work hardening in the plastic-deformation region
2.1. Stress analysis
by Jeong et al.10 The cross-section distortion formula was
deduced based on virtual force principle by Liu et al.11 It is
noted that, though many factors cannot be considered such According to the plane strain assumption and volume constant
as contact conditions and unequal stress/strain distributions, condition of plastic deformation, a strain field in tube bending
the analytical models are still very useful to estimate and pre- can be determined by
dict forming quality for tube bending. 8


>
> e ¼ ln Rþr cos u
¼ ln q
Considering the complexity of tube bending forming, finite < h Re Re

element (FE) numerical simulation method has been widely eu ¼ 0 ð1Þ

>
> t0 
used to explore the bending deformation under various forming :
et ¼ ln t ¼ eh
conditions. Fang et al.12,13 established a three-dimensional FE
model for high-strength 21-6-9 stainless steel tube in NC bend- where eh, eu, and et are tangential strain, hoop (circumferential)
ing and revealed the effects of mandrel on wall thinning and strain, and normal (thickness) strain, respectively; R is the cen-
cross-section distortion. By FE analysis, Li et al.14 addressed terline bending radius, Re curvature radius of neutral layer, q
the side push and booster push on wall thinning, cross- curvature radius of arbitrary fiber layer in tube wall, r the dis-
section distortion, and wrinkling occurring in NC bending of tance from any point in the cross-section to bending center,
thin-walled aluminum alloy tubes. Murata et al.15 numerically and u the span angle; t and t0 are wall thickness of tube before
studied the effects of strain hardening exponent on wall thick- and after bending, respectively (see Fig. 1). a is the angle of
ness variation, cross-section distortion and springback during neutral layer deviation.
tube NC bending. Yang et al.16,17 investigated the effects of The stress-strain relation satisfies the total strain theory if
friction on wall thinning, cross-section distortion and wrinkling the tube bending deformation satisfies the proportional load-
of Al-alloy and stainless steel tubes in NC bending by using the ing condition (assumption (1)) and it can be obtained as20
FE method. In Ref. 18, the coupling effects of the material 8
e
  
< eh ¼ r
rh  2 ru þ rt 
> 1
parameters and the bending angle on springback for TA18 tube
eu ¼ r

e ru  12 ðrh þ rt Þ ð2Þ
NC bending were researched by using FE simulation software >
:
e
  
ABAQUS. The springback mechanism for a large-diameter et ¼ r
rt  2 rh þ ru
1

thin-walled CT20 titanium alloy tube in NC bending was

where rh, ru and rt are tangential stress, hoop stress, and nor-
revealed using numerical simulation method by Zhan et al.19
and
e denote equivalent stress and
mal stress, respectively; r
Although a large number of researches have been carried
equivalent strain, respectively.
out to accelerate the development of tube bending technology,
By substituting Eq. (1) into Eq. (2), the normal stress rt can
the theory of tube bending mechanism needs to be further stud-
be deduced as
ied because of the limitations of forming theories and the defi-
ciency of stress test technologies. Thus, in this paper, some rt ¼ 2ru  rh ð3Þ
approximate formulae are derived to quantitatively predict wall Under the plane strain condition, equivalent stress r
and
thickness variation, cross-section distortion and springback equivalent strain
e can be obtained as20
based on plane strain assumption and virtual work principle qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1  2  2 pffiffiffi
in tube NC bending. The results are validated by the experi-
¼ pffiffiffi
r rh  ru þ ru  rt þ ðrt  rh Þ2 ¼ 3 rh  ru
mental results or validated simulation results in literature. 2
ð4Þ
2. Plastic deformation analysis of tube NC bending pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2  2  2 2

e ¼ eh  eu þ eu  et þ ðet  eh Þ2 ¼ pffiffiffi jeh j ð5Þ
3 3
According to the plastic deformation theory and the geometri-
cal feature of tube NC bending, the analytical models of tube By substituting Eqs. (1), (4) and (5) into the exponent strain
bending are proposed in terms of stress distribution, neutral hardening law r
¼ K
en , the relationship between tangential
layer position, bending moment, wall thickness variation, stress and hoop stress can be deduced as
1438 S. Lu et al.

Fig. 1 Schematic diagram of bent tube.

nþ1 n  pffiffiffi1þn
1 2 q ð1  cos uÞðK=2Þ 2= 3 ½ lnðq=Re Þn
rh  ru ¼  pffiffiffi K  ln ð6Þ rui ¼  ð15Þ
2 3 Re 2m þ cos u
When q > Re , the plus sign ‘‘+” is taken, otherwise the where ruo and rui denote hoop stress on the outer side and
minus sign ‘‘” is taken. inner side of the tube wall, respectively.
Stress equilibrium differential equation in hoop direction Substituting Eqs. (12)–(15) into Eq. (3), the normal stress rt
can be expressed as (the detailed derivation process is shown is calculated as
in Appendix A)  pffiffiffi1þn
ð1  2 cos u  2mÞðK=2Þ 2= 3 ½lnðq=Re Þn
rh r sin u rto ¼ ð16Þ
dru ¼ du ð7Þ 2m þ cos u
R þ r cos u
 pffiffiffi1þn
Integrating Eq. (7), the hoop stress ru can be obtained as ð1  2 cos u  2mÞðK=2Þ 2= 3 ½ lnðq=Re Þn
Z u rti ¼  ð17Þ
rh r sin u R þ r cos u 2m þ cos u
ru ¼ du ¼ rh ln ð8Þ
0 R þ r cos u Rþr where rto and rti denote normal stress on the outer side and
By using r = D/2 and m = R/D, where D is initial outer inner side of the tube wall, respectively.
diameter of tube, the hoop stress ru can be further written as
2.2. Neutral layer deviation
2mr þ r cos u 2m þ cos u
ru ¼ rh ln ¼ rh ln ð9Þ
2mr þ r 2m þ 1
The curvature radius of neutral layer for any cross-section can
The range of m in this paper should be great than or equal be expressed as21
to 2 to make the error smaller (the detailed discussion process
is shown in Appendix B). When the m P 2, the logarithmic A
Re ¼ R dA
ð18Þ
part of Eq. (9) can be simplified as A q

2m þ cos u 1  cos u 1  cos u where A denotes the area of cross-section, and

ln ¼ ln 1   ð10Þ
2m þ 1 2m þ 1 2m þ 1 A ¼ pD2 =4  pðD  2tÞ2 =4 is taken for the tube.
Therefore, the relationship between tangential stress and From the Eq. (18), the curvature radius of neutral layer Re
hoop stress can be expressed as can be obtained as

1  cos u A pD2 =4  pðD  2tÞ2 =4

ru ¼ rh ð11Þ Re ¼ R dA
¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2m þ 1 A q p 4R2  ðD  2tÞ2  4R2  D2
Substituting Eq. (11) into Eq. (6), the tangential stress rh ð19Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
can be deduced as 4R2  ðD  2tÞ2 þ 4R2  D2
pffiffiffi 1þn ¼
ð2m þ 1ÞðK=2Þð2= 3Þ ½ln ðq=Re Þn 4
rho ¼ ð12Þ
2m þ cos u Since D  t, Eq. (19) can be simplified as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi 1þn D2
ð2m þ 1ÞðK=2Þð2= 3Þ ½ lnðq=Re Þn Re ¼ R2  ð20Þ
rhi ¼  ð13Þ 4
2m þ cos u
where rho and rhi denote tangential stress on the outer side and The amount of neutral layer deviation En can be obtained
inner side of the tube wall, respectively. as En ¼ R  Re , while En in literature6 is calculated as
Incorporating Eqs. (11)–(13), the hoop stress ru is pre- 0:21D2
sented as En ¼ ð21Þ
R
 pffiffiffi1þn
ð1  cos uÞðK=2Þ 2= 3 ½lnðq=Re Þn The relationship among Re, R and the angle of neutral layer
ruo ¼ ð14Þ deviation a (shown in Fig. 1) can be expressed as
2m þ cos u
Plastic deformation analysis and forming quality prediction of tube NC bending 1439

Re ¼ R þ r cos a ð22Þ 0
Dto ¼ ttt o ¼ Dttout
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 R t 2 2tð1cos aÞffi ð27Þ
Substituting Eq. (20) into Eq. (22), the angle of neutral
¼ 2t=D
1 1
þ DR þ Dt  þDþD 
layer deviation a can be calculated as 2 2 D

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
R2  D2 =4  R
a ¼ arccos Dti ¼
t0i t
¼ Dttin
D=2 t
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 1 R 2t2 2t    1 R 2t
¼ arccos ð2R=DÞ2  1  2R=D ð23Þ ¼ 2t=D
1
 2 þ D þ D þ D 1 þ cos a  2t D
 2 þ D þ D

ð28Þ
2.3. Bending moment calculation where t0o and t0i denote the wall thickness of the extrados and
the intrados after bending, respectively; Dtout and Dtin denote
The distance between tensile/compressive fiber of tube and the amount of wall thickness thinning and thickening, respec-
neutral layer is y. y is positive value for tensile fiber, while y tively (Fig. 3).
is negative value for compressive fiber. In the neutral layer As can be seen from Eqs. (27) and (28), the wall thickness
position, y is zero as shown in Fig. 2. variation is related to the relative bending radius R/D, the rel-
When 0 6 u 6 2p, the y can be obtained as ative wall thickness of tube t/D and the angle of neutral layer
deviation a.
y ¼ rðcos u  cos aÞ ð24Þ
And the value range of y can be determined by 3.2. Cross-section distortion
rð1  cos aÞ 6 y 6 rð1  cos aÞ ð25Þ
Then, the bending moment M can be expressed as During tube NC bending process, tangential tensile stress sub-
Z Z Z jected to the outer material of bent tube will lead to wall thick-
2p D=2 2p
M¼ rh ydS ¼ rh yrdudr ness thinning, while tangential compressive stress subjected to
"Z
0 d=2 0 the inner material of bent tube will lead to wall thickness thick-
Z a
D=2 ening. Meanwhile both the resultant force of tangential stress
¼2 rho ðcos u  cos aÞr2 dudr
d=2 0
of inner part and that of outer part point to the center of
R D=2 R p i cross-section of tube, which leads to the cross-section distor-
þ rhi ðcos  cos aÞr2 dudr
d=2 a tion as shown in Fig. 3, where F denotes the tangential tensile
Z a
K 2
1þn
 3  ½lnðq=Re Þn force and M the bending moment.
¼ ð2m þ 1Þ pffiffiffi D  d3 ðcos u  cos aÞdu
24 3 0 2m þ cos u Based on the plane strain assumption, the hoop deforma-
Z p
½ lnðq=Re Þ n
tion of the bent tube is neglected. The tube is constrained in
 ðcos u  cos aÞdu
a 2m þ cos u transverse direction (major axis) by bending die groove and
ð26Þ under free deformation conditions in vertical direction (minor
axis) in bending process. Thus, the changing ratio of minor
where d is the inner diameter of tube. axis is often adopted to describe the severity of cross-section
distortion in engineering application.
3. Forming quality prediction of tube NC bending It is assumed that the wall thickness changes uniformly
along the axial direction of the tube and hoop deformation
3.1. Wall thickness variation is not considered. The changing ratio of outer diameter of
minor axis DD can be expressed as
Based on the plane strain assumption and considering neutral D  D0 Dtout Dtout =t
layer deviation, the wall thinning ratio Dto and wall thickening DD ¼ ¼ ¼ ð29Þ
D D D=t
ratio Dti can be derived as (the detailed derivation process is
shown in Appendix C) where D0 denotes the minimum outer diameter of minor axis
for the distorted cross-section.
Substituting Eq. (27) into Eq. (29), DD can be further
obtained as
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi3
2
1 41 R t 1 R t 2tð1  cos aÞ5
DD ¼ þ þ  þ þ 
2 2 D D 2 D D D

ð30Þ
Similarly

d  d0 Dtin Dtin Dtin =t

Dd ¼ ¼ ¼ ¼ ð31Þ
d d D  2t D=t  2

where Dd denotes the changing ratio of inner diameter of

minor axis and d0 the minimum inner diameter of minor axis
Fig. 2 Distance between tangential fiber and neutral layer. for the distorted cross-section.
1440 S. Lu et al.

Fig. 3 Cross-section distortion in bending process under plane strain condition.

Substituting Eq. (28) into Eq. (31), Dd can be further The virtual state equation of bent tube springback based on
expressed as virtual force principle can be expressed as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 1 R 2t2 2t    1 R 2t
 2 þ D þ D þ D 1 þ cos a  2t  2 þ D þ D MT ¼ 1 ð34Þ
D
Dd ¼
2  4t=D
MS ¼ 1 ð35Þ
ð32Þ
According to the virtual force principle, the springback
As can be seen from Eqs. (30) and (32) that the cross-
angle Dh can be calculated as
section distortion is also related to the relative bending radius
R/D, the relative wall thickness of tube t/D and the angle of Z h Z h
MT MS Mu Re Re M Re M
neutral layer deviation a. Dh ¼ dh ¼ dh ¼ h ð36Þ
0 EI 0 EI EI

3.3. Springback where E denotes elasticity modulus; I is the moment of inertia

of cross-section after bending, and the value is 0.7 times of that
Bent tube can be considered as statically determinate structure before bending.22
before unloading. One end of the bent tube is regarded as The moment of inertia of cross-section after bending can be
fixed, and the other is free when unloading occurs momentar- calculated as
ily. The reverse loading bending moment Mu is equal to the p 4 
loading bending moment M, but the direction is opposite. I ¼ 0:7  D  d4 ð37Þ
64
Springback is a question of small disturbance of elastic defor-
mation, which can be calculated by the virtual work principle. Substituting Eqs. (20), (26) and (37) into Eq. (36), Dh can be
The reverse bending moment Mu is loaded on free end of further obtained as
bent tube, and the virtual unit bending moment MT and MS pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffi 1þn
3:81Kh R2 D2 =4ð2mþ1Þð2= 3Þ ðD3 d3 Þ
are loaded on sectional T and S, respectively. Sectional T Dh ¼ 4
pEðD d Þ
4 
and S are assumed as the cross-section of the starting bending 2 n
point and ending bending point of tube, respectively; h is a Z ln pRþr cos u
ffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
bending angle, as shown in Fig. 4. The relative rotation angle 6 a
R D =4
2 2
6 ðcos u  cos aÞdu
of sectional T and S can be regarded as springback angle of the 4 2m þ cos u
0
ð38Þ
bent tube.
The actual state equation of bent tube springback can be n 3
expressed as Z  ln pRþr
ffiffiffiffiffiffiffiffiffiffiffiffiffi
cos u
ffi
p 2 R D =4
2 7
Mu ¼ M ð33Þ  ðcos u  cos aÞdu7
5
a 2m þ cos u

The length of the neutral layer remains unchanged before

and after springback, thus the curvature radius of neutral layer
after springback can be calculated as
Re h Re h
Rf ¼ ¼ ð39Þ
h0 h  Dh
where Rf denotes the curvature radius of neutral layer after
springback; h0 is the actual forming angle.
Then, the radius growth DR after springback can be
obtained as
DR ¼ Rf  Re ð40Þ
Fig. 4 Virtual state for bent tube springback.
Plastic deformation analysis and forming quality prediction of tube NC bending 1441

4. Evaluation of theoretical results

Based on the neutral axis deviation theoretical analysis, a cal-

culation example was carried out. The parameters of the exam-
ple are as follows:
The outside diameter of tube D is 10 mm, the wall thickness
of the tube t is 1 mm and relative bending radius R/D is 2.5.
The neutral axis deviation En in the paper is
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
D2
En ¼ R  Re ¼ R  R2  ¼ 0:51 mm
4
While the neutral axis deviation En in literature6 is
0:21D2 Fig. 6 Calculation and validated simulation results for cross-
En ¼ ¼ 0:84 mm
R section distortion degree.
For thin-walled tube bending with small diameter under
non-mandrel condition and the relative bending radius
R/D > 2, the neutral axis deviation En is less than 0.8 mm.23
Therefore, the neutral axis deviation formula in the paper is
better than that in Ref. 6.
Fig. 5 shows the comparison of wall thickening ratio
between calculation and experimental results. The 1Cr18Ni9Ti
stainless steel tube NC bending with the specification of
28 mm  1 mm (D  t) was carried out by the NC tube bender
VB300HP in literature.24 It can be seen from Fig. 5 that the
change trend of theoretical calculation results of wall thicken-
ing ratio is consistent with that of experimental results.
Namely the wall thickening ratio decreases with the increase
of relative bending radius. Compared with the experimental
results, the average relative error and maximum relative error
of the theoretical results of Ref. 8 are 31.21% and 35.30%, Fig. 7 Calculation and experimental results for springback
respectively; the average relative error and maximum relative angle.
error of the theoretical results of Ref. 6 are 29.87% and
35.96%, respectively; while the average relative error and max- cross-section distortion degree. It can be seen from Fig. 6 that
imum relative error of the theoretical results of the paper are the change trend of analytical calculation results of cross-
14.97% and 17.88%, respectively. As a whole, the analytical section distortion degree is consistent with that of simulation
calculation results of the paper are closer to the experimental results. Viz. the cross-section distortion degree decreases with
results when compared to the results of literature,6,8 which the increase of relative bending radius. Compared with the
can fully satisfy the project requirements for the fast predica- simulation results, the average relative error and maximum rel-
tion of the forming quality defects of tube bending. This is ative error of the analytical results of Ref. 8 are 87.87% and
because the neutral layer shift is considered in the paper, while 103.71%, respectively; the average relative error and maximum
that is neglected in the literature.6,8 relative error of the analytical results of Ref. 6 are 53.54% and
Fig. 6 shows a comparison between calculation results and 57.83%, respectively; while the average relative error and max-
simulation results validated by experiment in literature25 for imum relative error of the analytical results of the paper are
22.78% and 33.39%, respectively. By the large, the analytical
calculation results of the paper compared with that of litera-
ture6,8 is closer to the validated simulation results. This is also
because the neutral layer shift is considered in the paper, while
that is not considered in the literature.6,8
Fig. 7 shows a comparison of springback angle between cal-
culation and experimental results of 21-6-9 high strength stain-
less steel tube NC bending with the specification of
6.35 mm  0.41 mm  20 mm (D  t  R) in literature.26 It is
discovered that the change trend of the analytical calculation
results of springback angle is consistent with that of the exper-
imental results. Viz. with the increase of bending angle, the
springback angle increases. Compared with the experimental
results, the average relative error of analytical results of the lit-
erature27,28 is 44.90%, the maximum relative error of 68.02%
Fig. 5 Calculation and experimental results for wall thickening is located at the bending angle of 30°; the average relative error
ratio. of analytical results of the literature29 is 61.77%, the maximum
1442 S. Lu et al.

relative error of 77.82% is also located at the bending angle of dh dh

N ¼ rh rdudr sin þ ðrh þ drh Þrdudr sin ðA1Þ
30°; while the average relative error of the analytical results of 2 2
the paper is 25.00% and the maximum relative error of 56.48% For the small angles here take
is also located at the bending angle of 30°. Furthermore, when
the bending angle is larger than 90°, the relative error is less dh dh
sin  ðA2Þ
than 15.00%. On the whole, the analytical calculation results 2 2
of the paper compared with that of literature27–29 is closer to Substituting Eq. (A2) into Eq. (A1) and neglecting
the experimental results. This is because that the 3D stress/ the product of infinitesimal, the Eq. (A1) can be simplified
strain, moment of inertia of cross-section variation and neutral as
layer deviation are considered in the paper, while the 3D stress/
N ¼ rh rdudrdh ðA3Þ
strain is neglected, and the material is simplified into the ideal
elastic–plastic material in literature;27,28 the moment of inertia Stress equilibrium differential equation in hoop direction
of cross-section variation and neutral layer deviation are not can be expressed as
considered in literature.29
ru ðR þ r cos uÞdhdr þ N sin u  ðru þ dru ÞðR þ r cos uÞdhdr ¼ 0
The error is larger at smaller bending angle in the paper.
This is because that the smaller plastic deformation zone, ðA4Þ
material inhomogeneity, workpiece shape, and dimension Substituting Eq. (A3) into Eq. (A4), Eq. (A4) can be simpli-
inconformity lead to the larger impact on the small bending fied as
angle.28 The effect of the axis force on springback is neglected,
rh rdudrdh sin u  dru ðR þ r cos uÞdhdr ¼ 0 ðA5Þ
and the frictional condition and multi-die dynamic constraint
are not considered in tube bending process, which will intro- Therefore, the differential relationship between hoop stress
duce errors for the analytical results. and tangential stress can be obtained as
Through the above analysis, it is known that the precision
rh r sin u
of the analytical model in this paper is higher than that of the dru ¼ du ðA6Þ
literature. Thus, the model can be used to quickly predict the R þ r cos u
forming quality of tube NC bending.
Appendix B
5. Conclusions
As
(1) Based on the plane strain assumption and exponent
hardening law, the approximate formulae of stress, cur- fðxÞ ¼ lnð1 þ xÞ ðB1Þ
vature radius of neutral layer, angle of neutral layer
deviation and bending moment of tube NC bending f2n ðxÞ ¼ ð2n  1Þ!ð1 þ xÞ2n 06n ðB2Þ
are derived.
(2) Wall thinning ratio, wall thickening ratio, changing ratio f2nþ1 ðxÞ ¼ ð2nÞ!ð1 þ xÞð2nþ1Þ 06n ðB3Þ
of outer diameter of minor axis and changing ratio of
inner diameter of minor axis of tube NC bending are The Taylor development is expressed as follows:
deduced under the plane strain assumption and consid-
x2 x3 1 1
ering neutral layer deviation. lnð1 þ xÞ ¼ x  þ þ . . .  x2n þ x2nþ1 þ . . .
2 3! 2n 2n þ 1
(3) The springback angle of bent tube is derived according
to the virtual work principle, and the springback radius ðB4Þ
is derived on condition that the length of the neutral To make lnð1 þ xÞ  x hold, one condition must be satis-
layer remains unchanged before and after springback. fied, that is
(4) The analytical models are validated by the experimental
results or the simulation results validated in literature, x2
which can be used to quickly predict the forming quality jxj  ) 2  jxj ðB5Þ
2
of tube NC bending.
For example
10% error may request:
Acknowledgements
jxj x2 1
P ) jxj 6 ðB6Þ
The authors would like to thank the National Natural Science 10 2 5
Foundation of China (No. 51164030) and National Defense In Eq. (10)
Key Disciplines Laboratory of Light Alloy Processing Science
and Technology, Nanchang Hangkong University (No. 1  cos u 1
jxj ¼ 6 ðm > 0Þ ðB7Þ
gf201501001) for the support on this research. 2m þ 1 5

) m P 4:5 ðB8Þ
Appendix A
20% error may request:
The differential element body stress analysis of bent tube is jxj x2 2
shown in Fig. A1. The centripetal composite force of tangen- P ) jxj 6 ðB9Þ
5 2 5
tial stress N is
Plastic deformation analysis and forming quality prediction of tube NC bending 1443

σϕ+dσϕ
ϕ
σθ +dσθ dϕ σϕ+dσ
σθ
N dr
σϕ σϕ
N
r dϕ

R
R
dθ
O
Fig. A1 Differential element body stress analysis of bent tube.

1  cos u 2 Dti ¼
t0i t
¼ Dttin
jxj ¼ 6 ðm > 0Þ ðB10Þ t
2m þ 1 5 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 1 R 2t2 2t    1 R 2t
¼ 2t=D
1
 2 þ D þ D þ D 1 þ cos a  2t D
 2 þ D þ D
) mP2 ðB11Þ
ðC5Þ
30% error may request:
where t0i
denotes the wall thickness of the intrados after
3jxj x2 3
P ) jxj 6 ðB12Þ bending.
10 2 5

1  cos u 3 References
jxj ¼ 6 ðm > 0Þ ðB13Þ
2m þ 1 5
1. Ahmetoglu M, Altan T. Tube hydroforming: state-of-the-art and
) m P 1:17 ðB14Þ future trends. J Mater Process Technol 2000;98(1):25–33.
2. Yang H, Li H, Zhang ZY, Zhan M, Liu J, Li GJ. Advances and
trends on tube bending forming technologies. Chin J Aeronaut
Appendix C 2012;25(1):1–12.
3. Wu WY, Zhang P, Zeng XQ, Li J, Yao SS, Luo AA. Bendability
of the wrought magnesium alloy AM30 tubes using a rotary draw
According to eh = et, eu = 0 and considering neutral layer
bender. Mater Sci Eng, A 2008;486(1–2):596–601.
deviation, the equation can be obtained as 4. Liu KX, Liu YL, Yang H, Zhao GY. Experimental study on
Re þ rð1  cos aÞ  ðt  t0o Þ t0 cross-section distortion of thin-walled rectangular 3A21 alu-
ln ¼  ln o ðC1Þ minium alloy tube by rotary draw bending. Int J Mater Prod
Re t
Technol 2011;42(1–2):110–20.
where t0o denotes the wall thickness of the extrados after 5. Liu KX, Liu YL, Yang H. Experimental study on the effect of dies
bending. on wall thickness distribution in NC bending of thin-walled
Taking the logarithm for Eq. (C1) and substituting rectangular 3A21 aluminum alloy tube. Int J Adv Manuf Technol
Dtout ¼ t  t0o into Eq. (C1), Eq. (C1) can be simplified as 2013;68(5–8):1867–74.
6. Tang NC. Plastic-deformation analysis in tube bending. Int J Pres
Dt2out  ½Re þ rð1  cos aÞ þ tDtout þ rtð1  cos aÞ ¼ 0 ðC2Þ Vessels Pip 2000;77(12):751–9.
7. Al-Qureshi HA, Russo A. Spring-back and residual stresses in
Substituting r = D/2 into Eq. (C2), Dtout can be obtained as bending of thin-walled aluminium tubes. Mater Des 2002;23
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi3 (2):217–22.
2
D 41 R t 1 R t 2tð1  cos aÞ5 8. E DX, Chen JS, Ding J, Bai X. In-plane strain solution of stress
Dtout ¼ þ þ  þ þ 
2 2 D D 2 D D D and defects of tube bending with exponential hardening law. Mech
Based Des Struct Mach: Int J 2012;40(3):257–76.
ðC3Þ 9. Li H, Yang H, Song FF, Zhan M, Li GJ. Springback character-
ization and behaviors of high-strength Ti-3Al-2.5V tube in cold
Therefore, the wall thinning ratio Dto can be derived as
rotary draw bending. J Mater Process Technol 2012;212
0
Dto ¼ ttt o ¼ Dttout (9):1973–87.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 R t 2 2tð1cos aÞffi ðC4Þ 10. Jeong HS, Ha MY, Cho JR. Theoretical and FE analysis for
¼ 2t=D
1 1
2
þ DR þ Dt  2
þDþD  D
inconel 625 fine tube bending to predict springback. Int J Precis
Eng Manuf 2012;13(12):2143–8.
11. Liu JY, Tang CT, Ning RX. Deformation calculation of cross
Similarly, the wall thickening ratio Dti can be determined by
section based on virtual force in thin-walled tube bending process.
the following equation Chin J Mech Eng 2009;22(5):696–701.
1444
12. Fang J, Lu SQ, Wang KL, Yao ZJ. Three-dimensional finite
element model of high strength 21-6-9 stainless steel tube in rotary
draw bending and its application. Indian J Eng Mater Sci 2015;22
(2):142–51.
13. Fang J, Lu SQ, Wang KL, Xu JM, Xu XM, Yao ZJ. Effect of
mandrel on cross section quality in numerical control bending
process of stainless steel 2169 small diameter tube. Adv Mater Sci
Eng 2013;2013:1–9.
14. Li H, Yang H, Zhan M, Kou YL. Deformation behaviors of thin-
walled tube in rotary draw bending under push assistant loading
conditions. J Mater Process Technol 2010;210(1):143–58.
15. Murata M, Kuboki T, Takahashi K, Goodarzi M, Jin Y. Effect of
hardening exponent on tube bending. J Mater Process Technol
2008;201(1):189–92.
16. Yang H, Gu RJ, Zhan M, Li H. Effect of frictions on cross section
quality of thin-walled tube NC bending. Trans Nonferrous Met
Soc China 2006;16(4):878–86.
17. Yang H, Li H, Zhan M. Friction role in bending behaviors of thin-
walled tube in rotary-draw-bending under small bending radii. J
Mater Process Technol 2010;210(15):2273–84.
18. Jiang ZQ, Yang H, Zhan M, Xu XD, Li GJ. Coupling effects of
material properties and the bending angle on the springabck angle
of a titanium alloy tube during numerically controlled bending.
Mater Des 2010;31(4):2001–10.
19. Zhan M, Zhai HD, Yang H. Springback mechanism and
compensation of cryogenic Ti alloy tube after numerically
controlled bending. Trans Nonferrous Met Soc China 2012;22
(Suppl. 2):s287–93.
20. Yu HQ, Chen JD. Principle of metal plastic forming. Beijing: China
Machine Press; 2004. p. 107–10 [Chinese].
21. Liu HW. Mechanics of materials. 5th ed. Beijing: Higher Educa-
tion Press; 2011. p. 152–9 [Chinese].
22. Wang WC, Stelson KA. Computer aided manufacturing for three-
dimensional tube bending with on-line springback compensation.
Trans NAMRI/SME 1991;70–6.
23. E DX, Guo XD, Ning RX. Analysis of strain neutral layer
displacement in tube-bending process. J Mech Eng 2009;45
(3):307–10 [Chinese].
S. Lu et al.
24. Cheng PZ. Research on quality forecast and parameters opti-
mization technology of NC tube bending [dissertation]. Beijing:
Beijing Institute of Technology; 2010 [Chinese].
25. Fang J, Lu SQ, Wang KL, Xu XM, Xu JM, Yao ZJ. FE analysis
of section distortion in numerical control bending of the
0Cr21Ni6Mn9N stainless steel tube. J Plast Eng 2013;20(5):71–6
[Chinese].
26. Fang J, Lu SQ, Wang KL, Yao ZJ. Springback law of high
strength 21-6-9 stainless steel tube in NC bending under different
process parameters. Proc IMechE Part B-J Eng Manuf 2015;26
(3):379–84.
27. E DX, He HH, Liu XY, Ning RX. Spring-back deformation in
tube bending. Int J Min Met Mater 2009;16(2):177–83.
28. Liu JW, Ni XD, Gao WG. Modeling and analyzing the bending
re-spring angle of pipe Q235-A. Modern Manuf Eng 2003(5);51–2
[Chinese].
29. Zhang LL. Calculation of springback in pipe bending. Forg Stamp
Technol 2002(3);37–9 [Chinese].
Lu Shiqiang is a professor and Ph.D. supervisor at School of Aero-
nautical Manufacturing Engineering, Nanchang Hangkong Univer-
sity. He received the Ph.D. degree from Northwestern Polytechnical
University. His main research interests are plastic deformation theory
and technology of difficult to deform materials, finite element simu-
lation of plastic forming process.
Fang Jun received Ph.D. degree from Nanjing University of Aero-
nautics and Astronautics in 2015. He is now a teacher in Jiangxi Sci-
ence & Technology Normal University. His main research interests are
advanced plastic forming and multi-scale modeling & simulation.
Wang Kelu is a professor and Ph.D. supervisor with Ph.D. degree. His
main research interest lies in precision plastic forming theory and
technology.